Exfoliative cytology as a diagnostic test for basal cell carcinoma: a meta-analysis

Background Exfoliative cytology is only occasionally used in clinical practice to diagnose basal cell carcinoma (BCC). Objectives To systematically review the literature examining exfoliative cytology as a diagnostic tool for BCC and to meta-analytically summarize the accuracy of this test. Methods Diagnostic test meta-analysis. A computerized database search was made of MEDLINE (1966-2000) and EMBASE (1950-2000) using appropriately indexed terms. Hand searches of relevant journals were made. Bibliographies of relevant articles were further explored. Histopathology was used as the reference standard. A summary receiver-operating characteristic curve analysis was performed using the statistical package Meta-Test version 0.6 (J. Lau, New England Medical Center, Boston, MA, U.S.A., 1997). Results Eight primary studies (seven English language and one Italian language) fulfilled the inclusion criteria. The pooled sample included 1261 BCCs. These studies varied greatly in methodological quality and were, in general, poor. A meta-analysis showed the pooled sensitivity to be high at 97%[95% confidence interval (CI) 94-99]. Consequently, only 3% of BCCs included were misdiagnosed as non-BCC by cytology. The corresponding pooled specificity was 86% (95% CI 80-91). Conclusions Exfoliative cytology probably has a high diagnostic accuracy for BCC. However, large, better designed and better reported studies are needed to ascertain the true accuracy of this technique. In the interim, this test should be considered in specific subgroups of patients in whom even a 2-mm punch biopsy may be considered inappropriate, e.g. a cosmetically sensitive site in a young person. Similarly, it should be considered when a BCC is to be treated without a diagnostic biopsy being taken, e.g. with cryotherapy.